Charity Effectiveness and Third-World Economics

post by jimrandomh · 2013-06-12T15:50:22.330Z · LW · GW · Legacy · 37 comments

In a recent Facebook status update, Eliezer Yudkowsky asked a question:

Does the causal model for GiveDirectly's positive effects imply that the government of those countries could achieve the same effects by printing money in the local currency and giving the same amount to the same recipients? "Yes" is a legitimate answer because it's a dreadful truth that many governments around the world are not increasing their money supply enough, and also that choosing the right recipients can redistribute value productively even when supplies of medium-of-exchange are already sufficient.

My first thought was object-level; the obvious answer is that some fraction of the money given will eventually be converted into imports, transferring the burden of inflation out and onto richer countries which can easily afford it. This seems plausible. If true, it implies that we should multiply our effectiveness estimates by dImports/d$, which is (asspull) 0.5. By this line of reasoning, direct giving is less effective than we thought, but still a reasonably good deal.

My second thought was that it's likely true that some developing country governments could improve their economies by printing and distributing money, but they won't because they're corrupt, and giving directly is a workaround to force that policy upon them. This seems plausible at first, but it feels forced; the leaders' incentives here are ambiguous, not clearly aligned against this sort of policy.

My third thought was that it's likely true that developing countries' governments could improve their economies by printing and distributing money, and they might not know this.

Sanity check. What sort of people do the poorest countries' governments have, in their economic advisory roles? Is anyone making a serious effort to connect good economists with governments that need them?

If developing countries are short on competent economic advisors at the top levels, and no one is working to fix this, then funding that charity would outperform direct giving by multiple orders of magnitude. But what reason do we have to think that a well-placed economist can make a difference? Well, history does contain at least one big, salient success story: Brazil, where a clever scheme halted hyperinflation and turned the economy around. And on a smaller scale, Otjivero-Namibia.

So now I have some questions for the efficient altruism community:

 - Which developing nations have competent economic advisors, and which ones need them?
 - If a developing nation's leader needs good economic advisors to fill his/her cabinet, does he/she get them?
 - Do any nations have economic problems that seem especially amenable to fixing by clever economists?

37 comments

Comments sorted by top scores.

comment by DanielLC · 2013-06-12T23:21:46.720Z · LW(p) · GW(p)

If the government shouldn't be printing money and giving it to the poor, doesn't that imply that they should be taking money from the poor and destroying it? Or did they somehow manage to find exactly the right amount of money for the poor to have?

Replies from: Viliam_Bur, Alicorn
comment by Viliam_Bur · 2013-06-13T07:46:01.311Z · LW(p) · GW(p)

Giving money to people of taking money from people would have exactly the opposite effects if people wouldn't think and plan their future.

If people think about their future, know what the government typically does, and include it in their plans, then not giving or taking (too much) money allows people to plan long-term, which can help them live better.

Also, too much giving or taking money invites people to negative-sum fights about which group should be given most money, and which group should be taken from.

Replies from: DanielLC
comment by DanielLC · 2013-06-13T23:33:19.976Z · LW(p) · GW(p)

If you're suddenly giving vs. suddenly taking, either suggest a chaotic government, so people will react similarly. However, if the government did either consistently, then the effects would be opposite.

comment by Alicorn · 2013-06-13T00:23:22.430Z · LW(p) · GW(p)

Cute, but the results of the reversal test doesn't necessarily imply that any specific intervention ought to take place to correct an incorrect value.

Replies from: DanielLC
comment by DanielLC · 2013-06-13T00:37:47.251Z · LW(p) · GW(p)

It doesn't imply that it's the best use of your money, but it shows that the idea that something like that would help isn't surprising.

comment by paulfchristiano · 2013-06-13T11:10:12.637Z · LW(p) · GW(p)

If any charity is efficient (or merely more efficient than burning the money), then according to the philanthropist's values it would be better for the world if any government printed money and spent it on that charity. I don't think GiveDirectly is different in any meaningful way. As far as I can tell it sometimes seems different to people because thinking about giving cash reminds them that economics is a thing, which they often forget when thinking about other interventions.

There are some more subtle issues with inflation expectations, but I think those are really second-order considerations compared to the obvious "society gets more of what you pay for, and slightly less of everything else."

comment by elharo · 2013-06-13T10:33:29.159Z · LW(p) · GW(p)

Why are we only talking about developing nations? If the last 6 years show us anything, it's that developed nations more often than not don't have competent economic advisors; and aren't listening to the ones they have. In particular, developed nations have mostly failed to follow the policy of monetary growth and inflationary spending that was appropriate for their situations.

I see no evidence that economists in developing nations are any less competent or wise than economists in developed nations. In fact, this demonstrates a very common fallacy: do not assume that people in the developing world are less smart or rational than people in the developed world. A much more accurate model of the world is formed by assuming that people everywhere have similar distributions of intelligence and rationality, and rationally respond to different circumstances based on their perceived self-interest and limited understanding. This doesn't mean they are perfectly rational (in the economic, not LessWrong sense of the word "rational) actors, any more than people in the developed world are. However it does mean that what you see on the ground is more likely to be a result of serving the self-interest of some individuals and groups than a simple failure to understand the situation economically.

Furthermore, developed nations that manage their own currency are much better able to inflate their way out of a liquidity trap. A nation like Panama that does not have its own currency, or a nation like Costa Rica whose currency is pegged to the U.S. dollar, or a nation like Cuba whose currency isn't freely exchangeable, or any nation whose debt is denominated in units other than that nation's own currency is much less able to implement such a strategy effectively. Greece's problems have been massively exacerbated by the Euro. If they had stayed with the Drachma they'd be in better shape. Contrast Greece with Iceland which had similar debt issues but maintained its own currency, and weathered the financial crises much more easily. A nation that with its own hard currency is much better prepared to redistribute wealth in this fashion. If its debts are denominated in that currency, so much the better.

comment by Vaniver · 2013-06-12T16:24:50.130Z · LW(p) · GW(p)

My second thought was that it's likely true that some developing country governments could improve their economies by printing and distributing money, but they won't because they're corrupt, and giving directly is a workaround to force that policy upon them. This seems plausible at first, but it feels forced; the leaders' incentives here are ambiguous, not clearly aligned against this sort of policy.

If I remember correctly, didn't a significant portion of Zimbabwe's funding come from some Scandinavian charity which, as official policy, gave out local currency rather than dollars, and required buying the local currency from the government with real money at the official rate?

comment by Vaniver · 2013-06-12T16:35:07.944Z · LW(p) · GW(p)

Is anyone making a serious effort to connect good economists with governments that need them?

Hopefully someone else will remember the actual source for this, as I can't find it quickly, but I believe the World Bank provides economic advice to African countries, which includes gems like "sell off commodity rights by public, transparent auction, rather than having the minister decide which country gets the rights," which will lead to billions more in the country's coffers (instead of millions more in the minister's coffers).

I think the "print money and hand it out" plan is a terrible one. If it replaced price caps on basic foodstuffs, that would be better as a starvation prevention measure (since it would encourage more production, rather than less production).

I think a better path for economic advice from the developed world is the Hernando de Soto path; see his organization or his book.

Replies from: jimrandomh
comment by jimrandomh · 2013-06-12T17:10:41.245Z · LW(p) · GW(p)

I think the "print money and hand it out" plan is a terrible one. If it replaced price caps on basic foodstuffs, that would be better as a starvation prevention measure (since it would encourage more production, rather than less production).

Why do you think it's terrible?

Replies from: Vaniver
comment by Vaniver · 2013-06-12T19:00:16.664Z · LW(p) · GW(p)

Why do you think it's terrible?

Because inflation works as a tax on wealth- and the trouble with the undeveloped world is that there's too little wealth, and in particular, too little incentive for many people to generate formal wealth. Basic income guarantees have their merits, but funding them by inflation instead of a tax on exports or land or so on just seems misguided.

Replies from: RolfAndreassen, DanielLC, Eliezer_Yudkowsky
comment by RolfAndreassen · 2013-06-13T03:28:09.182Z · LW(p) · GW(p)

Because inflation works as a tax on wealth

No, it's a tax on currency-denominated wealth, which is not the same thing at all. Wealth in the form of land, houses, mining rights, guns, loyal soldiers, gold, cows, or friends at court is not taxed by inflation. To the extent that the problem in poor countries is that a small elite controls most of the cash economy - I do not make a claim on how large an extent this is - inflation would indeed solve it, provided that the tax extracted was given to the poor. Which is somewhat unlikely, but there you go.

Replies from: Vaniver
comment by Vaniver · 2013-06-13T03:41:15.897Z · LW(p) · GW(p)

No, it's a tax on currency-denominated wealth

I'm aware; I strongly suspect that it is much easier for the poor (that would be receiving this help) to store their wealth in the local currency than any of the forms you listed.

The issue is more that the small elite controls the legal economy, with most of the economy occurring off the books, leading to very little in the way for legal protection of the more sophisticated economic organizations which help people accumulate wealth.

(de Soto tells the story of being a government economist in Peru, where they were all worried that the construction market was collapsing because of declining revenues and permits. But something seemed odd- there were still cranes and buildings going up all over Lima, and concrete sales were seeing steady growth. They scratched deeper and discovered that extra-legal construction, which they thought was a few percent of legal construction, turned out to be significantly larger- the majority of the construction economy in the country was off the books.)

Replies from: RolfAndreassen
comment by RolfAndreassen · 2013-06-13T16:56:50.604Z · LW(p) · GW(p)

I strongly suspect that it is much easier for the poor (that would be receiving this help) to store their wealth in the local currency than any of the forms you listed.

Ok, fair enough. As noted, I don't make any claim on the extent to which the problem consists of cash being mainly in elite hands. That said, if you were willing to start with a blank slate, you might be able to redistribute with a one-time inflation event. Making up numbers: Suppose a 10% elite controls 90% of the wealth in both currency and non-currency forms, and suppose that these two are roughly equal. Do a 100% devaluation of the currency, ie full hyperinflation, redistributing the gains equally. The 10% still control 90% of the non-currency wealth, but now they only have 10% of the currency wealth. Presumably there is then some re-equilibration and you end up with some of the non-currency wealth making its way into 90% hands in exchange for the new currency.

Of course, in practice it is the wealthy who would be able to protect themselves from your scheme by moving their currency holdings into dollars, or whatever; and I've no idea how realistic my distribution between currency and non-currency is. If 90% of wealth is in non-currency form then you're hacking about at the edges. In any case, though, a single hyperinflation event is probably a bit different from the steady-but-low inflation in the original question.

comment by DanielLC · 2013-06-12T23:18:09.288Z · LW(p) · GW(p)

and in particular, too little incentive for many people to generate formal wealth.

I'm skeptical of that. The undeveloped world has less welfare (since they have less money to give) which means people are more desperate to make money.

I suppose it's possible that the problem is that, since they tend to be much closer to the brink of starvation, their less willing to take risks for money. Better to have a sure subsistence wage than a risky luxury wage. In that case, giving the poor more of the wealth would seem to help.

Replies from: Vaniver
comment by Vaniver · 2013-06-12T23:28:47.248Z · LW(p) · GW(p)

I'm skeptical of that.

I strongly recommend reading The Mystery of Capital. Its basic premise is that most developing countries have very weak legal systems, which dramatically reduces the level of trust individuals can have with each other, leading to dramatically lower incentives for wealth accumulation. Why bother improving your tin shack when your neighbors could collectively agree that it's not your shack and kick you out, or the President's cousin who officially owns the land you're squatting on could bulldoze it at any time? Why enter into a business agreement with someone else, when you can't enforce any written agreement between the two of you? Much safer to invest only with your family members who will have a much harder time running away with the profits, though they may not be the best people for the job.

The answers are also not found in modern American or European law, but in frontier American law; things like homesteading make great sense in most of the developing world, but are rarely implemented, often because of ignorance that it's a good or possible option.

Replies from: DanielLC
comment by DanielLC · 2013-06-13T00:47:00.270Z · LW(p) · GW(p)

That's an entirely separate issue. If you want people to make contracts, you don't do it by making them desperate. You do it by making them keep their contracts.

Replies from: Vaniver
comment by Vaniver · 2013-06-13T02:10:33.593Z · LW(p) · GW(p)

That's an entirely separate issue. If you want people to make contracts, you don't do it by making them desperate. You do it by making them keep their contracts.

The claim is that it's not an entirely separate issue, because bad legal systems are (in de Soto's view, at least) the primary cause of poverty in the third world, and so fixing the legal systems will seriously reduce the poverty.

The phrase "too little incentive" may have been unclear; I meant it in the sense that income taxes reduce the incentive to put effort into earning income, not that the utilons that result from post-tax earnings are lower. A lack of legal protection for capital held by the poor makes it often irrational for the poor to invest heavily in capital that they are not sure will remain theirs, and so they remain poor. Adding legal protections changes the incentive structure, and thus changes rational behavior.

Replies from: DanielLC
comment by DanielLC · 2013-06-13T02:16:57.547Z · LW(p) · GW(p)

And what does that have to do with printing money and giving it to the poor?

Replies from: Vaniver
comment by Vaniver · 2013-06-13T02:44:30.381Z · LW(p) · GW(p)

Claim: Primary controllable cause of poverty in the developing world is low incentive for the poor to develop formal wealth.

Claim: Printing money lowers the incentive to develop formal wealth because it is a tax on formal wealth (in cash form, at least).

Conclusion: Printing money would exacerbate primary controllable cause of poverty in the developing world.

Now, this is not the total cost-benefit analysis, but it is the part that will dominate it given my values and discount function.

Replies from: jimrandomh
comment by jimrandomh · 2013-06-13T13:16:08.047Z · LW(p) · GW(p)

Claim: Primary controllable cause of poverty in the developing world is low incentive for the poor to develop formal wealth.

No, this is absurd. Poor people do not respond to complex tax incentives; in most cases they don't have the education or information to even comprehend them. The primary causes of poverty in the developing world are lacks of resources of various kinds (capital, education, nutrition), which are themselves caused by past poverty, in a self-reinforcing cycle.

Replies from: Vaniver
comment by Vaniver · 2013-06-13T18:25:14.609Z · LW(p) · GW(p)

Poor people do not respond to complex tax incentives; in most cases they don't have the education or information to even comprehend them.

Tax incentives aren't the incentives I'm talking about here. I'll point you to de Soto again, as he makes the argument much more convincingly than I can.

The primary causes of poverty in the developing world are lacks of resources of various kinds (capital, education, nutrition)

This agrees with de Soto's view, but the resource he focuses on is "trust," which can be developed by developing the legal systems.

comment by Eliezer Yudkowsky (Eliezer_Yudkowsky) · 2013-06-12T19:34:05.307Z · LW(p) · GW(p)

The reason why I asked my original question is that GiveDirectly is directly handing money to the recipients, and this money is presumably competing with other money to make purchases. If one were to naively measure the effectiveness via the direct observation, "Ooh, the people with more money are better off than the people with less money!" then this kinda begs the question. It's not obvious to me that foreign money behaves differently in this regard if it's competing for goods that other low-income communities are also trying to purchase. Again, you can justify this model but it's not as simple as observing that the people with more money are better off. You need the country to be running an NGDP deficit / aggregate demand shortfall, for purchased imports to help, for redistribution of purchasing power to be good, etc.

Replies from: bogus, Vaniver
comment by bogus · 2013-06-12T20:10:11.991Z · LW(p) · GW(p)

The reason why I asked my original question is that GiveDirectly is directly handing money to the recipients, ...

What do you mean by "money" here? Local currency? They bought that currency in the foreign exchange market, so no, there's no injection of cash into the broader economy. The only issue I can think of is that spending a lot of money (here meaning generally "resources") in a very localized area might drive up local prices (i.e. land, local wages etc.) compared to the surroundings, making the locals not as much better off in PPP (purchasing-power parity) terms. But this is going to be negligible in a typical intervention.

Replies from: Eliezer_Yudkowsky
comment by Eliezer Yudkowsky (Eliezer_Yudkowsky) · 2013-06-12T20:13:35.140Z · LW(p) · GW(p)

Who would have otherwise bought that currency in the foreign exchange market? Where would they have spent it? Are higher prices on the foreign exchange market a good thing?

comment by Vaniver · 2013-06-12T20:34:26.356Z · LW(p) · GW(p)

Oh, I agree with you that GiveDirectly has second and third-order effects that may be more significant than the obvious effect of "we handed them cash!", which I why I prefer to view interventions in terms of how they shift incentives rather than their immediate visible impact. (This is the primary reason why I strongly prefer interventions that set up or alter institutions, rather than transfer resources. Unfortunately, directly exporting good governance has gone out of style, and so instead we have resource extraction firms exporting good resource extraction ability but little else.)

comment by Randaly · 2013-06-12T23:39:20.053Z · LW(p) · GW(p)

My third thought was that it's likely true that developing countries' governments could improve their economies by printing and distributing money, and they might not know this.

Sanity check. What sort of people do the poorest countries' governments have, in their economic advisory roles? Is anyone making a serious effort to connect good economists with governments that need them?

I'm fairly confident that Eliezer's perspective is that [most] developed countries' governments can also [currently] improve their economies by printing and distributing money, given that he's a fan of market monetarism.

comment by Larks · 2013-06-12T22:01:44.206Z · LW(p) · GW(p)

Giving out cash to voters sounds like exactly the sort of policy third world countries are very good at doing. In fact, the very worst countries like Zimbabwe seem to be the best at it.

Replies from: jimrandomh
comment by jimrandomh · 2013-06-12T22:34:48.792Z · LW(p) · GW(p)

According to Wikipedia's page on hyperinflation in Zimbabwe, they printed money to finance "involvement in the Democratic Republic of the Congo and, in 2000, in the Second Congo War, including higher salaries for army and government officials", "self-dealing", and "institutionalized corruption". In other words, they were printing money and not giving it to voters.

Replies from: Vaniver
comment by Vaniver · 2013-06-13T02:25:50.655Z · LW(p) · GW(p)

In other words, they were printing money and not giving it to voters.

This is Zimbabwe we're talking about; I think they know who the real voters are.

Replies from: Multiheaded
comment by Multiheaded · 2013-06-13T07:52:33.720Z · LW(p) · GW(p)

Don't shift definitions in mid-argument.

Replies from: Vaniver
comment by Vaniver · 2013-06-13T18:28:16.616Z · LW(p) · GW(p)

I'm not Larks, but it seemed reasonable to me that by "voters" they meant "people with political power" instead of "peasants." But my view of Africa is more cynical than normal, and perhaps that wasn't the intended definition, in which case I retract the grandparent.

comment by Luke_A_Somers · 2013-06-12T16:14:58.923Z · LW(p) · GW(p)

we should multiply our effectiveness estimates by dImports/d$, which is (asspull) 0.5

Don't you mean (1 - d(Imports)/d$)?

And of course even if all of it went to buying imported goods some benefit would still accrue. If those were high-grade capital goods, it could be better than 100% domestic spending!

Replies from: jimrandomh, jkaufman
comment by jimrandomh · 2013-06-12T17:04:45.600Z · LW(p) · GW(p)

No, I mean d(Imports)/d$. There are two things that make this tricky. The first trick is that money is not wealth; you have to track the flow and creation of goods, not the flow of money. The second trick is that money is not a conserved quantity; governments can create and destroy it at will.

Consider a hypothetical country which is a closed system (ie, it has no imports and no exports), and suppose you take money from outside and give it to someone in that country. This has two effects on the recipient country. The first effect is that that person will use the money to buy things, and making them wealthier. This is a positive effect. The second effect is that they bid up the price of those things in the recipient country, so everyone else has to pay very slightly more. This manifests as a slight increase in inflation, which is a negative effect. In a well-functioning economy with zero income inequality, these effects are exactly equal, but one is measurable and the other is hidden. There are two ways around this.

There are two reasons why the direct benefit would be larger than the diffuse harm. The first is income inequality: when you give money to very poor people, they spend it more productively than richer people, because richer people will have already used up their highest-value buying opportunities. In the best case, giving people money lets them transform themselves from starving idle peasants into a productive working class, which goes on to create value (goods) worth much more than the initial handout. But this particular gain does not require the money to come from outside; the local government can simply print money and hand it out, and unlike an outside donor, they can do so for free. The only cost they pay is inflation, which they suffer in both cases.

The second way around the inflation problem is through imports. This is where governments can't just print money, because the international trade value of their local currency falls faster than the local trade value. If you give people in a country money and that money is eventually spent on imports, then the inflation burden spreads through international markets, and mostly lands on rich countries, where it is much less harmful.

Replies from: Luke_A_Somers
comment by Luke_A_Somers · 2013-06-12T19:01:45.192Z · LW(p) · GW(p)

The first trick is that money is not wealth; you have to track the flow and creation of goods, not the flow of money.

Yes, and that's in specific why I was making this argument. This money isn't being dumped into the economy as a whole - it's targeted to poor families. Now, money is the lubricant of the economy, and those folks were jammed tight: since they're so short on cash, it's tough for them to maintain an economy. They need to do it by barter or in their heads. So if this greases the wheels and lets them begin trading with each other, then you get really serious gains - people able to work in ways they weren't before. If this money leaves, that's bad news indeed.

Now, secondly, consider: these folk are stuck on the bottom relative to their neighbors. Effectively, you can split them off as a sub-Nigeria with no economic policy at all. When they buy things from their wealthier neighbors, that fulfils the role of the export that you describe above.

Basically, I think your second and third thoughts in the OP are the dominant issues.

comment by jefftk (jkaufman) · 2013-06-12T18:39:13.322Z · LW(p) · GW(p)

Luckily he guessed 0.5, so it doesn't matter!

comment by CronoDAS · 2013-06-14T07:11:05.801Z · LW(p) · GW(p)

One effect of inflation is that it tends to transfer wealth from creditors to debtors. Poor people are more likely to be debtors than creditors, so inflation would tend to work to their advantage. There was once a U.S. presidential candidate whose primary campaign promise was to cause inflation by increasing the money supply...